The MCP server ships free with the open-source Trigger.dev CLI. It operates against your Trigger.dev account; the platform has a free tier plus paid plans billed on usage. You pay only for the runs and compute you consume.

Does it support remote/OAuth?

It runs locally over stdio via npx trigger.dev@latest mcp and authenticates with a personal access token (TRIGGER_ACCESS_TOKEN) rather than OAuth. Flags such as --dev-only and --readonly let you restrict the session to the dev environment or hide write tools.

Trigger.dev MCP server

OfficialTrigger.devConfig last verified Jun 1, 2026

Trigger.dev's official MCP server: scaffold projects, trigger and debug background tasks, deploy to any environment, and run TRQL queries from your agent.

Trigger.dev is a platform for long-running background tasks and durable workflows that you write in TypeScript and run without managing queues or timeouts. Its official MCP server ships inside the Trigger.dev CLI, so an agent can add Trigger.dev to a project, trigger and debug task runs, deploy to production or preview environments, and query operational data — turning the assistant into a hands-on operator of your background jobs.

The server exposes a broad tool surface across the whole lifecycle: project and org management (list orgs and projects, create a project, initialize Trigger.dev in a codebase), tasks (list tasks, inspect the current worker, fetch a task's payload schema, trigger a task), runs (list and filter runs, get run and span trace details, wait for a run to complete, cancel a run), deployments and preview branches, a TRQL query engine for analyzing runs and metrics with companion schema and dashboard tools, a local dev-server controller, CLI profile management, documentation search, and a managed-prompts toolkit plus chat-agent tools for tasks built with chat.agent(). It is launched with npx trigger.dev@latest mcp over stdio and authenticates with a personal access token (TRIGGER_ACCESS_TOKEN); flags like --project-ref, --dev-only, --readonly, --api-url, and --log-level scope and harden the session. The search_docs tool works without authentication.

Quick install

Copy-paste configs are provided for all 8 supported clients. Pick your client below.

Add to ~/.claude.json

~/.claude.json

json

{
  "mcpServers": {
    "trigger-dev": {
      "command": "npx",
      "args": [
        "-y",
        "trigger.dev@latest",
        "mcp"
      ],
      "env": {
        "TRIGGER_ACCESS_TOKEN": "<TRIGGER_ACCESS_TOKEN>"
      }
    }
  }
}

Or via CLI

bash

claude mcp add trigger-dev -- npx -y trigger.dev@latest mcp

Step-by-step guides

Available tools

Tool	Description
search_docs	Search across the Trigger.dev documentation to find relevant information, code examples, API references, and guides. Works without authentication.
list_projects	List all projects for the current user; useful when searching for a project or looking up a projectRef.
list_orgs	List all organizations for the current user. Useful when looking up an org slug or ID.
create_project_in_org	Create a new project in an organization. Use only when adding Trigger.dev to an existing project.
initialize_project	Initialize Trigger.dev in your project, creating a new project in the organization you select.
get_tasks	Get all tasks in the project; useful when searching for a task or looking up a task identifier/slug.
get_current_worker	Get the current worker for the project, including version and registered task slugs.
get_task_schema	Get the payload schema for a specific task. Use get_current_worker first to see available task slugs.
trigger_task	Trigger a task in the project. Use get_tasks to list tasks and ask the user to select one.
get_run_details	Get the details and trace of a run; trace events are paginated, with the first call returning run details and the first page.
get_span_details	Get detailed information about a specific span within a run trace.
wait_for_run_to_complete	Wait for a run (ID starting with run_) to complete, with an optional timeoutInSeconds parameter (default 60s).
cancel_run	Cancel a run by its ID, which starts with run_.
list_runs	List all runs for a project; use when searching for a run or listing all runs.
deploy	Deploy a project, triggering a deployment. This is a long-running operation.
list_deploys	List all deploys for a project; use when searching for a deploy or listing all deploys.
list_preview_branches	List all preview branches for a project.
query	Execute a TRQL query against your Trigger.dev data — a SQL-style query language for analyzing runs and metrics.
get_query_schema	Get the column schema for a specific TRQL table. Available tables: runs, metrics, llm_metrics.
list_dashboards	List available built-in dashboards with their widgets. Each dashboard contains pre-built queries for common metrics.
run_dashboard_query	Execute a single widget query from a built-in dashboard. Use list_dashboards first to see what's available.
whoami	Show the current authenticated user, active CLI profile, email, and API URL.
list_profiles	List all configured CLI profiles and show which one is currently active.
switch_profile	Switch the active CLI profile for this MCP session, changing which Trigger.dev account is used.
start_dev_server	Start the Trigger.dev dev server (trigger dev) in the background, waiting up to 30 seconds for the worker.
stop_dev_server	Stop the running Trigger.dev dev server.
dev_server_status	Check the status of the dev server and view recent output (stopped, starting, or ready).
list_prompts	List all managed prompts in the project environment, showing slug, current version, and override status.
get_prompt_versions	List all versions for a specific prompt, showing version number, labels, source, model, and content.
promote_prompt_version	Promote a prompt version to be the current active version. Only code-sourced versions can be promoted.
create_prompt_override	Create a dashboard override for a prompt that takes precedence over the current code version.
update_prompt_override	Update the active dashboard override for a prompt. Works only if an override is currently active.
remove_prompt_override	Remove the active dashboard override for a prompt, reverting to the current code version.
reactivate_prompt_override	Reactivate a previous dashboard-sourced version as the active override.
list_agents	List all chat agents in the current worker — tasks created with chat.agent() or chat.customAgent().
start_agent_chat	Start a conversation with a chat agent. Returns a chatId for use with send_agent_message.
send_agent_message	Send a message to an active agent chat and get the full response text back; the agent remembers full context.
close_agent_chat	Close an agent chat conversation so the agent exits its loop gracefully.

Required configuration

TRIGGER_ACCESS_TOKENRequired
Trigger.dev personal access token (tr_pat_...). Authenticates the MCP server. Obtained via trigger.dev login or the dashboard.

What you can do with it

Trigger and debug background tasks from your editor

Ask the agent to run a task with a test payload: it calls get_tasks, get_task_schema, then trigger_task, and follows up with get_run_details and get_span_details to walk the trace when something fails — all without opening the dashboard.

Deploy and analyze workloads conversationally

Have the agent deploy to a preview branch with deploy, watch the result, then run a TRQL query to surface error rates or slow runs over the last hour using query and run_dashboard_query.

FAQ

Is it free?: The MCP server ships free with the open-source Trigger.dev CLI. It operates against your Trigger.dev account; the platform has a free tier plus paid plans billed on usage. You pay only for the runs and compute you consume.
Does it support remote/OAuth?: It runs locally over stdio via npx trigger.dev@latest mcp and authenticates with a personal access token (TRIGGER_ACCESS_TOKEN) rather than OAuth. Flags such as --dev-only and --readonly let you restrict the session to the dev environment or hide write tools.

← Browse all automation servers